Re: [PATCH -mm] tracing: Fix section mismatch in trace_hw_branches.c

[Frederic Weisbecker <fweisbec@gmail.com>]

On Sun, Feb 22, 2009 at 07:33:08PM +0600, Rakib Mullick wrote:
> On 2/21/09, Frederic Weisbecker <fweisbec@gmail.com> wrote:
> 
> > Hi,
> >
> >  When I saw this patch, I searched the real purpose of __cpuinit and its
> >  real impact.
> >  But I didn't find any comments about it inside the kernel.
> >
> >  But today, by looking at the discussion around latest git pull for x86
> >  to mainline, I discover that __cpuinit becomes __init on UP.
> >
> >  So, unless I missed something, this patch seems to me very dangerous.
> >  The init and reset callbacks of a tracer can be called at any time, not only
> >  on initcalls time (__init functions are freed from memory after the middle stage
> >  of the boot).
> >  With this patch, on UP we will dereference freed memory while activating this tracer.
> If the init and reset callbacks of a tracer can be called regardless
> of cpu hotpluging then it is. If the tracer's init or reset doesn't
> rely on cpuhotplug then it shouldn't use it.
> There's a another way to fix the warning is by remove __cpuinitdata
> from bts_hotcpu_notifier.


Yes, they can be called on UP, on SMP with or without cpu hotplug,
and everytime (boot, runtime).

init() is called when you switch to a tracer:

echo tracer_name > /debug/tracing/current_tracer

and reset() is called when you switch to another one.

But removing __cpuinitdata will mean a kind of waste of memory (though it's only
a little struct).

 
> Thanks,